We have developed a method for isolation, expansion, and selection of tumor-reactive T lymphocytes from either solid ovarian cancer or malignant ascites. Routinely, freshly isolated T cells are placed in anti-CD3 monoclonal antibody-coated tissue culture flasks for 48 hours. Subsequently, T cells are reactivated every 7-10 days for three times with irradiated antologous tumor cells to select for tumor-reactive T cells. After the final reactivation, T cells are tested for recognition of autologous and allogeneic tumor cells. These studies have shown that tumor-reactive T cells can indeed be expanded from almost all specimens. They also showed that therecognition of tumor cells is HLA class I-restricted and T cell receptor mediated. Further studies showed that in particular HLA-A2 is a dominant molecule in presenting tumor antigens. Since HLA-A2 is expressed by about 50% of the population, we decided to focus on HLA-A2-restricted tumor antigens. A more recent finding was that HLA-A2-matched tumors are better recognized by BLA-A2-restricted T cells than HLA-A2-negative tumors. This suggested the presence of shared tumor antigens. One of these was recently identified by us as a nine amino acid peptide derived from Her2/neu, a protooncogene that is overexpressed in 20-30% of ovarian and breast cancers. Using the now commonly used method of acid elution of HLA-bound peptides on tumor ceRs and HPLC fractionation, we have preliminary evidence that multiple shared tumor antigens exist in both ovarian and breast cancer. Thus the methodology that was earlier applied to melanoma appeared also effective in other malignancies. Therefore, we have selected tumor cell lines that express these antigens, but do not overexpress Her2/neu, to identify the nature of these common antigens. We hope that with mass spectrometry we will be able to do so. We are presently studying peptide fractions from both an ovarian and a breast cancer cell line.